Clays and Clay Minerals, Vol. 47. No, 6, 742-747, 1999.

D E T E R M I N A T I O N OF M E A N C R Y S T A L L I T E D I M E N S I O N S F R O M X - R A Y D I F F R A C T I O N P E A K PROFILES: A C O M P A R A T I V E A N A L Y S I S OF S Y N T H E T I C HEMATITES
MARIO CROSA,1 VALTER BOERO,t AND MARINELLA FRANCHINI-ANGELA 2 Dipartimento di Valorizzazione e Protezione delle Risorse Agroforestali-Chimica Agraria, Universit5 di Torino, via Leonardo da Vinci 44, I 10095 Grugliasco (Torino), Italy 2Dipartimento di Scienze Mineralogiche e Petrologiche, Universit/a di Torino, via Valperga Caluso 35, 1-10125 Torino, Italy Abstract--X-ray diffraction (XRD) profile analysis of eight synthetic hematite samples was performed to identify the best parameters for determining the apparent mean crystallite dimension (D) and, consequently, surface area. The samples are comparable to soil hematite with respect to crystallinity. The procedure included: a) deconvolution of the XRD peaks to Gauss and Cauchy components and subtraction of the instrumental profile, b) determination of D from full-width at half-maximum, integral breadth, and integral breadth measurements of the Cauchy component, and c) comparison of deduced surface areas with those obtained by the N2-BET adsorption method. As expected, D values are strongly influenced by the broadening parameters. An appropriate selection of peaks is required to obtain size values along the crystallographic axes a (hkl: 110, 300) and c (hkl: 104, 116) and to calculate reliable surface areas. Using the Cauchy component of the above peaks, the calculated surface areas compared well with those measured by the N2-BET adsorption method. Key Words-BET, Hematite, Iron Oxides, Mean Crystallite Dimension, Powder X-ray Diffraction, Surface Area, XRD. INTRODUCTION Hematite (et-Fe203, space group R3c) is a common high-resolution transmission electron microscopy is not satisfactory to obtain data about behavior (e,g., sorption) (Schwertmann, 1987). This is because soil is usually strongly clustered, and the habit of the constituent particles may be easily determined, but not their morphology. Furthermore, only few X-ray diffraction (XRD) peaks of soil hematite do not superpose over peaks from other minerals, thereby making analysis difficult. The XRD method appears to be the only practical and reliable way to describe the surface area, which requires the use of crystal-size measurements. Among suitable XRD parameters, peak broadening is currently used (e.g., Schwertmann and Latham, 1986) to determine the "mean crystallite dimension" (MCD) of hematite, i.e., the size of the coherent scattering domain perpendicular to the diffracting (hkl) plane (Schwertmann and Cornell, 1991), or more recently, the mean coherence length (MCL, Cornell and Schwertmann, 1996), which is equivalent to MCD. The purpose of this work is to investigate how best to evaluate the XRD profile and to determine how the choice of specific XRD reflections may influence the apparent crystal size. We then compare the specific surface area obtained by the method proposed herein and that obtained by the N2-adsorption method. THEORETICAL BACKGROUND The broadening of the XRD diffraction-peak profile h(x) of a microcrystalline substance is assumed to be a convolution of the pure profile f(x) and the instrumental profile g(x) (e.g., Klug and Alexander, 1974,

Fe(III) oxide in soils and sediments of warm climates

(Cornell and Schwertmann, 1996). Even at low (1-10 wt. %) concentrations, hematite may strongly affect chemical-physical properties such as matrix color (Torrent et al., 1983), sorption of anions (e.g., Torrent et al., 1994), organics (Schwertmann et al., 1986) and heavy metals (e.g., Pb, Cu, Zn, Co, Ni, and Mn; Schwertmann and Taylor, 1989), and interactions with phyllosilicates (e.g., soil structure, aggregation, and cementation; Schwertmann, 1987). These properties are strongly affected by surface area and this, in turn, by morphology, habit, and size of the hematite crystals. The "morphology" is defined as the assemblage of the forms present in the crystal without regard to the extent of face development (Hartmann, 1973, p. 369). The "habit" of a crystal depends on the relative development of each form present in the morphology of the crystal. Therefore, crystals having the same morphology, e.g., prism and bipyramid, may show different habits: prismatic habit, when the prism is the most developed form; bipyramidal habit, when the prism is negligible with respect to the well developed pyramid. Unfortunately, such crystal properties are very difficult to observe directly and to measure in the pedoenvironment because soil hematite consists of very small crystallites and is often present at low concentrations. The direct observation of hematite particles by using scanning electron microscopy, transmission electron microscopy (TEM), or Copyright 9 1999, The Clay MineralsSociety

742

Vol. 47, No. 6, 1999

180 160 140

XRD profile analysis and mean size of synthetic hematites

350
R 2 = 0.89

743

300 250 i200 -'%0

15o lOO 50 o
o

R2 = 0.52

i
-r

120
lOO

so

60 40 20
0

20

40

60
D ( f l ) (nm)

80

100

120

20

40

60
D ( f l ) (nm)

80

100

120

Figure 1. Relation between the mean sizes D calculated from FWHM and from 13.

Figure 2. Relation between the mean sizes D calculated from 13~and from 13.

p. 291). fix) originates from the microcrystallinity of the phase and g(x) is obtained from a reference phase free of strain and 2-20 txm in size. Experimental data of g(x) and h(x) are usually fitted with mathematical functions such as the pseudo-Voigt or the Pearson VII functions (Howard and Preston, 1989). Methods may be adopted also to deconvolute f(x) and consequently to determine the crystallite size. In a simplified way, assuming a negligible contribution of strain in fix), the mean size, D, can be calculated by the full-width at half-maximum (FWHM) peak characteristic or by the area/height XRD-peak ratio (integral breadth, 13), using the equation: D = K M13 cos 0 (1)

where K is the Scherrer constant, h the wavelength, 13 the F W H M (in radians), and 0 is the peak angular position. If strain and size components can be related to well-defined profile shape functions, f(x) may be regarded as the convolution of a Gauss function (strain component) and a Cauchy function (size component, e.g., Langford et al., 1987). Therefore, the integral breadth of the Cauchy component ([3c) can be used in Equation (1) instead of 13 to obtain D. Jones (1981) adopted this approach for soil hematite and, more recently, Colombo et al. (1994) did the same for synthetic hematite. Both [3 and [3c were determined here for comparison. FWHM, despite its debated physical

significance (Langford and Wilson, 1978) is used in the analysis of soil iron oxides, and the K constant is conventionally set to 0.9 (Scbwertmann and Cornell, 1991). With the use of integral breadths and the K constant set to unity, the equation allows the determination of an apparent size. In effect, because fix) is not independent of either the crystallite shape or size distribution (Wilson, 1949; GuErin and Alvarez, 1995), a proper value of constant K, which is a function of reflection and crystallite shape, must be adopted if a " t r u e " value of D is desired. If particle shape and size uniformity are assumed, K and D can be calcuhlted for several peaks by l e a s t - s q u a r e s fit, as adopted by Stanj ek (1991) for hematite. In soil samples, however, diffraction peaks overlap or overlap partially from other phases, which may reduce the number of hematite peaks of satisfactory quality. Thus, the information of crystallite shape and size as derived from the diffraction peaks is not easy to obtain. The most useful information about hematite crystallite size and shape may be obtained from Equation (1) and D by determining D a and D c values, i.e., mean crystallite size parallel to crystallographic unit lengths a and c. From these values, an estimation of the surface area in hematite may be obtained assuming a quasi-cylinder shape, even though it does not always rep-

Table 1. Da and Dc values (nm) of sample H1 obtained from different hkl and broadening parameters.
Peaks Paxameter 012 104 110 113 024 116 018 214 300

Da values

FWHM [3 13o FWHM [3 13c

9.6 7.6 10.2 12.2 9.7 13.0

12.2 9.9 13.8 15.5 12.5 17.5

29.3 22.5 27.7

21.1 17.3 25.6

Dc values

9.8 7.7 10.1 12.5 9.8 12.8

12.1 10.0 14.8 15.5 12.7 18.9

3.3 2.7 3.8 16.9 13.7 19.3

21.3 17.4 25.4 10.8 8.8 12.9

32.7 25.2 31.2

13.3 10.9 16.2

744

Crosa, Boero, and Franchini-Angela

Clays and Clay Minerals

Table 2. Da values (nm) of the hematite samples, obtained as average of nine peaks ~. The standard deviations are reported in parentheses.
Samples Parameter H1 H2a H2b H3 H4 H5 H6 H8 H 11

FWHM [3 [3~

16.8 (9.8) 13.4 (6.8) 18.1 (9.6)

46.0 (15.7) 40.0 (17.2) 10.9 (72.2)

28.8 (11.2) 20.3 (7.2) 16.3 (5.4)

34.7 (15.6) 28.1 (13.3) 39.9 (18.5)

37.7 (16.6) 29.7 (13.0) 39.1 (18.2)

21.9 (11.9) 16.3 (6.8) 18.5 (8.8)

40.2 (21.9) 30.3 (16.4) 35.6 (18.9)

47.8 (48.3) 31.3 (27.9) 22.8 (11.2)

19.8 (10.5) 15.1 (7.8) 18.2 (9.2)

1Eight peaks for samples H2a and H2b.

resent a close approximation to the actual habit (e.g., Torrent et al., 1987). M ATER I A L S A N D M E T H O D S

Synthesis o f hematite
To obtain reliable X R D and BET measurements, eight hematite samples, with comparable crystallinity to those of soil, were synthesized by various methods (Schwertmann and Cornell, 1991). Samples H I and H2 were obtained by hydrolysis of Fe(NO3)3 and Fe(C104) 3 solutions, respectively, and aged for 7 d at 98~ H3 and H4 were synthesized from FeC13 solutions in 0.002 and 0.001 M HCI, respectively, and aged for 10 d at 98~ Samples H5 and H8 were prepared by transformation of ferrihydrite in the presence of solutions of N a n C O 3 (2 d, 90~ and oxalic acid (36 h, 70~ solutions, respectively. Sample H8 was obtained by adding 40 mL 0.1 M AI(NO3)3 to 160 mL 0.1 M Fe(NO3)3, adjusting the pH to 7 --- 0.2, then washing, adjusting again to pH 7, and storing at 80~ for 70 d. Sample H11 was obtained similarly to sample H1, but aged 14 d. All products were washed with 1 N KOH, then with deionized water, dialyzed, and freeze-dried.

Self-supporting powder mounts were prepared by gently pressing 200 mg of sample into an A1 holder. The powder mounts were smoothed at >0.9 m m in thickness to obtain the highest diffraction intensity (Klug and Alexander, 1974). All samples were step scanned (25-80 ~ 0.02 ~ intervals, and 20 s counting time per increment) to include nine peaks: 012, 104, 110, 113, 024, 116, 018, 214, and 300. The digitized X RD profiles were fitted using a split pseudo-Voigt function (Howard and Preston, 1989), with a modified Q B A S I C program of Enzo et al. (1985), which included the separation of CoKc~ (0.17988 nm) components. Wherever possible, each peak was fitted independently and the Marquardt algorithm was used to obtain the best value of goodnessof-fit (GOF%): GOF% = 100[Z (Io - lc)2/(/o)2]1/2 (2)

where Io is the observed intensity and I c is the calculated intensity.

Data analysis
The instrumental profile, g(x), was obtained based on Stanjek (1991) from a reference sample of synthetic c o m m e r c i a l hematite (Merck, no. 3924, lot no. 548003) previously heated to 1000~ for 48 h. The 2 20-~m size fraction was obtained by sieving and by gravity sedimentation in water. An X R D scan of this size fraction was performed at 0.01 ~ interval and 80 s counting time.

XRD analysis
XR D was performed with a Philips diffractometer (PW1729 generator, PW1820 goniometer) equipped with 1~ diverging slit, 0.2-mm receiving slit, and 1~ scatter slit. The C o K a (0.179026 nm) radiation involved a system with an Fe filter at 40 kV and 40 mA.

Table 3. Dc values (nm) of the hematite samples, obtained as average of seven peaks 1. The standard deviations are reported in parentheses.
Samples Parameter H1 H2a H2b H3 H4 H5 H6 H8 H 11

FWHM [3 [3 c

13.8 (2.2) 11.2 (1.8) 15.8 (2.9)

26.8 (18.8) 38.1 (16.2) 87.l (17.0)

21.7 (11.2) 15.5 (7.2) 16.1 (5.4)

32.6 (7.1) 26.5 (5.9) 38.4 (10.0)

35.4 (7.3) 27.9 (5.8) 36.9 (8.5)

18.4 (3.2) 14.1 (2.9) 17.4 (5.4)

33.6 (5.2) 25.7 (4.4) 31.4 (7.9)

24.8 (1.3) 18.6 (1.1) 21.3 (2.8)

17.3 (3.2) 13.2 (2.3) 16.2 (2.7)

1 Six peaks for samples H2a and H2b.

Vol. 47, No. 6, 1999

XRD profile analysis and mean size of synthetic hematites Table 4. Da (nm) evaluated from [~ of selected peaks (100 and 300).
Samples

745

Peak

H1

H2a

H2b

H3

H4

H5

H6

H8

H 11

110 300 Mean

22.5 25.2 23.8

24.4 21.0 22.7

71.0 62.3 66.6

38.7 47.6 43.2

42.6 48.6 45.6

27.4 28.7 28.0

50.0 56.1 53.0

78.6 77.5 78.1

24.4 26.7 25.6

For all the samples and for each peak, the Voigt function was obtained from the pseudo-Voigt parameters for h(x) and g(x) of both K a I and K~xz. The Cauchy and the Gaussian component of the broadening were then separately deconvoluted, and the FWHM, [3, and [3c of the Voigt f(x) was obtained. Details of this procedure are in deKeijser et al. (1982) and Crosa (1996). The value of D, perpendicular to hkl planes, was obtained from the FWHM, 13, and [3c using Equation (1) with K = 0.9 when the value for F W H M was calculated, and K = 1 when the value of [3 or [3c was used. The thickness of the crystal (domain) along any crystallographic axis can be derived from D by multiplying by cosc~, where c~ is the angle between the hkl plane and the crystallographic direction under consideration (Schwertmann, 1987). Therefore, the values of D a and D c were computed from D and the related cosine function.
Specific s u r f a c e a r e a

maining. For a similar procedure, see Benedetti et al. (1988).

Influence o f b r o a d e n i n g p a r a m e t e r s o n D

The specific surface area was measured by N 2 adsorption on samples that were outgassed (at 60~ using a FISON Sorptomatic 9000 apparatus. Values were obtained with the BET isotherm and the software from the manufacturer. The estimation of the surface area from XRD data, by using a cylindrical shape approximation, was obtained from the formula of Torrent et al. (1987): Specific surface area (m2g 1) = 7 6 0 / D a + 3 8 0 / D c RESULTS AND DISCUSSION
Experimental profiles

From all reflections in all the samples, Figures 1 and 2 show the effect of FHHM, [3, or [3c on the D values. In general, these differences are caused by the nature of f(x) (Gauss/Cauchy ratio) and by the K values [Equation (1)]. In effect, in a Voigt profile (deKeijser et al. 1982), the FWHM/[3 ratio (the so-called Voigt parameter) may range between 0.6366 (pure Cauchy) and 0.9394 (pure Gaussian). Moreover, based on the K-constant values adopted, the ratio of D values obtained from FWHM and [3 may range from 0.9581 to 1.4138. Furthermore, in terms of D, the [3J[3 ratio may range from 0 to 1 and the FWHM/[3c ratio may vary consequently. In these hematite samples, a near linear regression exists between D obtained from FWHM vs. D obtained from [3 (R e = 0.97) and D obtained from [3~ vs. D from [3 (R 2 = 0.90), if the size is < 6 0 nm. Scattering increases strongly above the size limit (Figures 1 and 2).
D a a n d D c estimation

(3)

The GOF% of all the profiles was comparable. These values were judged satisfactory for all peaks in the samples (GOF%: 0.58-1.09). The reference sample showed a GOF% slightly higher than that of each hematite, particularly for the 012 reflection, a very asymmetric peak. Also, the 018 peak often showed an overlap by a spurious peak. Sample H2 showed negative values of the Gaussian component: each peak was therefore assumed to be the sum of two Gaussian components as derived from two populations of crystals, H2a and H2b, of different mean size. From preliminary TEM photographs, there is good agreement between XRD and TEM-size determinations. From peakintegrated intensities, H2a was estimated at 32.0(1)% by volume of the total sample and H2b for the re-

As an example, the values of D a and D c for sample HI, obtained from different peaks, are reported in Table 1. For all the samples, Tables 2 and 3 give, for comparison, the average parameters based on all reflections. Differences are evident for each peak and for the broadening parameter used. Using [3 as a broadening parameter, the average value of D a ranges (excluding sample H2a) from 13.4 to 31.3 nm. Because the standard deviation of D a in some of the samples is greater than this range and, considering also the limitations of the XRD method (to 100-200 nm) the error in the use of [3 as a broadening parameter may be high. Differences among reflections may be related to the following: 1) the D value is related to an apparent size rather than an actual size; 2) the estimation of D a from an hkl plane not perpendicular to a is dependent, at least partially, on the D c value and vice versa. Consequently, the best D a or D c estimation may be obtained from peaks with hkO-hO0 and 00l Miller indices, respectively. Therefore, suitable reflections for determining D a are 300 and 110. Table 4 shows for each sample the D a value (from [3) obtained from the above peaks. The crystallite shape (and also structural strain)

746

Crosa, Boero, and Franchini-Angela

Clays and Clay Minerals

1 0 0 ~ 9 0

80
60

~ 1 0 4 012 / 024

so high that any comparison is not significant (data not shown). CONCLUSIONS We conclude that: 1) the evaluation of apparent crystallite size D is clearly affected by the choice of the parameter used in the X R D analysis. 2) As expected, the reliability of D a and D c values depends on the use of an appropriate peak; in general only well-resolved peaks from planes perpendicular to crystallographic axes allow the determination of a reliable D a and D c value. Nevertheless, Dc, lacking any 001 peak, may be estimated from an appropriate Okl, hOl, or hkl peak, but only if D c figures so obtained are lower than their corresponding m ax i m u m value, which can be obtained from Da. The D a and D c values so obtained are only apparent sizes. However, especially for soil hematite, it is difficult to convert these values to actual size, but this is possible when both habit, and morphology of the (uniform) crystallites are known. Unfortunately, habit and morphology of hematite in a soil are not always obtainable. However, the use of apparent size does not appear to prevent a reliable surface area estimation. 3) The combination of proper peak selection with a "single-peak m e t h o d " (e.g., Cauchy-broadening parameter) to obtain the size effect from a peak profile, appears to be the best procedure to determine surface area values which closely fit those of the N2-BET method. 4) The use of F W H M and 13 can be applied rigorously, but only in the absence of strain, although these two broadening parameters are the appropriate parameters to use as demonstrated by others (e.g., Torrent et al., 1994). In fact, the F W H M parameter is commonly used probably because it is measurable even for low-quality X R D profiles, whereas ~ requires a higher quality peak profile and a fitting procedure. As noted by Borggaard (1990), the strain effect on soil iron oxides is unknown. Consequently, the application of F W H M or ~ may produce an incorrect evaluation of crystal size. Despite these potential problems, the adoption of a single-peak method is to be recommended, although it requires accurate data collection and sophisticated computation procedures.

10

20

30

40
Da

50 60 (nm)

70

80

90

100

Figure 3. Maximum measurable values for Dc estimation with the Okl, hOl and hkl peaks as a function of Da.

may explain residual differences between values obtained from the 110 and 300 reflections. In general, the use of the mean value for D a may be considered reliable. The D c estimation is more complex. Because peaks with 001 indices are unavailable, the estimates using Okl and D a values, and their effect on the D c estimation, may be calculated. Figure 3 shows the maximum value for Dc, as a function of Da. The 018 peak, which is the most favorable theoretically, does not appear as a well resolved reflection. Therefore, in the present work, the 116 and 104 peaks are used to estimate D c (Table 5). In addition, uncertainties about g(x) suggest the exclusion of the 012 reflection and the higher order 024 peak, both of which can be affected by strain. The D c values obtained from the 116 and 104 peaks are, for all hematite samples, always lower than the maximum values obtained from Figure 3 and, hence, underestimation of D c may be avoided. In each sample, the differences in these D c figures are probably due to shape and irregularities of the crystals. Specific surface area estimation Table 6 reports the specific surface areas calculated using XRD data and those by the N2-BET method. As expected, differences are evident with respect to various broadening expressions. The best agreement between the specific surface area values obtained by N2BET and XR D is observed when the 13c parameter is used. Obviously, when the specific surface area estimation is based on all X R D peaks, the differences are

Table 5. Dc (nm) evaluated from [3 of selected peaks (104 and 116).

Sarnples Peak H1 H2a H2b H3 H4 H5 H6 H8 H 11

116 104 Mean

12.7 12.5 12.6

14.3 18.3 22.7

45.0 49.0 47.0

32.4 32.3 32.4

33.4 34.7 34.1

18.2 16.4 17.3

32.0 29.5 30.8

19.3 20.0 19.7

15.2 15.0 15.1

Vol. 47, No. 6, 1999

XRD profile analysis and mean size of synthetic hematites

747

Table 6. Surface area (m z g-l) of hematites obtained from selected XRD peaks and the N2-BET method.
Samples Parameter H1 H2 t H3 H4 H5 H6 H8 HI 1

FWHM [~ 13~ BET-N 2

49.1 62.3 47.6 43.3

32.5 43.0 36.6 36.6

23.5 29.4 21.2 23.5

21.5 27.6 21.5 21.5

37.1 50.3 45.6 47.3

20.5 27.2 23.2 27.4

20.1 28.8 40.0 41.3

41.5 55.1 47A 48.0

Weighted mean of H2a and H2b.

ACKNOWLEDGMENTS The authors are grateful to M. Baricco, Dipartimento di Chimica I.EM., Universit~t di Torino, Italy, for helpful discussions, and the anonymous referees and S. Guggenheim for their important suggestions, and to M.U.R.S.T., Roma (Grant: "quota ex 4 0 % " ) for financial support. REFERENCES Benedetti, A., Fagherazzi, G., Enzo, S., and Battagliarin, M. (1988) A profile fitting procedure for analysis of broadened X-ray diffraction peaks. II. Application and discussion of methodology. Journal o f Applied Crystallography, 21, 543-549. Borggaard, O.R. (1990) Dissolution and Adsorption Properties o f Soil Iron Oxides. Chemistry Department, Royal Veterinary and Agricultural University, Copenhagen, 122 pp. Colombo, C., Barron, V., and Torrent, J. (1994) Phosphate adsorption and desorption in relation to morphology and crystal properties of synthetic hematites. Geochimica et Cosmochimica Acta, 58, 1261-1269. Cornell, R.M. and Schwertmann, U. (1996) The Iron Oxides. Verlagsgesellschaft, Weinheim, Germany, 573 pp. Crosa, M. (1996) Ottimizzazione dell'analisi dei profili XRD per la misura delle dimensioni medie dei cristalli di ematite e goethite, Ph.D. thesis, Universit~ degli Studi di Torino, Italia, 100 pp. deKeijser, Th.H., Langford, J.I., Mittemeijer, E.J., and Vogel, A. (1982) Use of the Voigt function in a single-peak method for the analysis of X-ray diffraction peak broadening. Journal o f Applied Crystallography, 15, 308-314. Enzo, S., Polizzi, S., and Benedetti, A. (1985) Applications of fitting techniques to the Warren-Averbach method for Xray peak broadening analysis. Zeitschriftfiir Kristattographie, 170~ 275-287. Guerin, D.M.A. and Alvarez, A.G. (1995) A survey on the determination of crystal size in powder diffractometry. Crystallography Review, 4, 261-281. Hartmann, P. (1973) Introduction to Crystal Growth. North Holland, Amsterdam, 470 pp. Howard, S.A. and Preston, K.D. (1989) Profile fitting of powder diffraction patterns. In Modern Powder Diffraction, Reviews in Mineralogy, Volume 20, D.L. Bish and J.E. Post, eds., Mineralogical Society of America, Washington, D.C., 217-275. Jones, R.C. (1981) X-ray diffraction peak profile analysis versus phosphorus sorption by eleven Puerto Rican soils. Soil Science Society of America Journal, 45, 818-825.

Klug, H.P. and Alexander, L.E. (1974) X-ray Diffraction Procedures f o r Polycrystalline and Amorphous Materials, 2nd edition, Wiley and Sons, New York, 966 pp. Langford, J.I. and Wilson, A.J.C. (1978) Scherrer after sixty years: A survey and some new results in the determination of crystallite size. Journal o f Applied Crystallography, 11, 102-113. Langford, J.I., Delhez, R., deKeijser, Th.H., and Mittemeijer, E.J. (1987) Profile analysis for microcrystal peak properties by the Fourier and other methods. Australian Journal o f Physics, 41, 173-187. Schwertmann, U. (1987) Some properties of soil and synthetic iron oxides. In Iron in Soil and Clay Minerals, J.W. Stucki, B.A. Goodman, U. Schwertmann, eds., NATO Advanced Study Institute, Series 217, Reidel Publishing Company, Dordrecht, Holland, 203-250. Schwertmann, U. and Cornell, R.M. (1991) Iron Oxides in the Laboratory: Preparation and Characterization. Verlagsgesellschaft, Weinheim, Germany. 137 pp. Schwertmann, U. and Latham, M. (I986) Properties of iron oxides in some New Caledonian oxisols. Geoderma, 39, 105-123. Schwertmann, U. and Taylor, R.M. (1989) Iron Oxides. In Minerals in Soil Environments, 2nd edition, J.B. Dixon and S.B. Weed, eds., Soil Science Society of America Book Series No. 1, Madison, Wisconsin, 379-438. Schwertmann, U., Kodama, H., and Fischer, W.R. (1986) Mutual interactions between organics and iron oxides. In Interactions o f Soils Minerals with Natural Organics and Microbes, P.M. Huang and M. Schnitzer, eds., Soil Science Society of America Special Publication No. 17, Madison, Wisconsin, 223-250. Stanjek, H. (1991) Aluminium- und Hydroxylsubstitution in synthetischen und natiirlichen Hfimatite. Ph.D. thesis, Buch am Erlbach, M~nchen, 194 pp. Torrent, J., Schwertmann, U., Fechter, H., and Alfrrez, E (1983) Quantitative relationship between soil colour and hematite content. Soil Science, 136, 354-358. Torrent, J., Schwertmann, U., and Barron, V. (1987) The reductive dissolution of synthetic goethite and hematite in dithionite. Clay Minerals, 22, 329-337. Torrent, J., Schwertmann, U., and Barron, V. (1994) Phosphate sorption by natural hematites. European Journal o f Soil Science, 45, 45-51. Wilson, A.J.C. (1949) X-ray Optics. Metuen and Co. Ltd., London, 127 pp. E-mail of corresponding author: boero@agraria.unito.it (Received 6 July 1998; accepted 16 May 1999; Ms. 98088)